Re: [PATCH net] net: vlan: fix a UAF in vlan_dev_real_dev()

On Wed, Oct 27, 2021 at 06:46:40PM -0700, Jakub Kicinski wrote:

On Wed, 27 Oct 2021 20:16:06 +0800 Ziyang Xuan wrote:

quoted

The real_dev of a vlan net_device may be freed after
unregister_vlan_dev(). Access the real_dev continually by
vlan_dev_real_dev() will trigger the UAF problem for the
real_dev like following:

==================================================================
BUG: KASAN: use-after-free in vlan_dev_real_dev+0xf9/0x120
Call Trace:
 kasan_report.cold+0x83/0xdf
 vlan_dev_real_dev+0xf9/0x120
 is_eth_port_of_netdev_filter.part.0+0xb1/0x2c0
 is_eth_port_of_netdev_filter+0x28/0x40
 ib_enum_roce_netdev+0x1a3/0x300
 ib_enum_all_roce_netdevs+0xc7/0x140
 netdevice_event_work_handler+0x9d/0x210
...

Freed by task 9288:
 kasan_save_stack+0x1b/0x40
 kasan_set_track+0x1c/0x30
 kasan_set_free_info+0x20/0x30
 __kasan_slab_free+0xfc/0x130
 slab_free_freelist_hook+0xdd/0x240
 kfree+0xe4/0x690
 kvfree+0x42/0x50
 device_release+0x9f/0x240
 kobject_put+0x1c8/0x530
 put_device+0x1b/0x30
 free_netdev+0x370/0x540
 ppp_destroy_interface+0x313/0x3d0
...

Set vlan->real_dev to NULL after dev_put(real_dev) in
unregister_vlan_dev(). Check real_dev is not NULL before
access it in vlan_dev_real_dev().

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Reported-by: <redacted>
Signed-off-by: Ziyang Xuan <redacted>
 net/8021q/vlan.c      | 1 +
 net/8021q/vlan_core.c | 2 +-
 2 files changed, 2 insertions(+), 1 deletion(-)

diff --git a/net/8021q/vlan.c b/net/8021q/vlan.c
index 55275ef9a31a..1106da84e725 100644
+++ b/net/8021q/vlan.c

@@ -126,6 +126,7 @@ void unregister_vlan_dev(struct net_device *dev, struct list_head *head)
 
 	/* Get rid of the vlan's reference to real_dev */
 	dev_put(real_dev);
+	vlan->real_dev = NULL;
 }
 
 int vlan_check_real_dev(struct net_device *real_dev,

diff --git a/net/8021q/vlan_core.c b/net/8021q/vlan_core.c
index 59bc13b5f14f..343f34479d8b 100644
+++ b/net/8021q/vlan_core.c

@@ -103,7 +103,7 @@ struct net_device *vlan_dev_real_dev(const struct net_device *dev)
 {
 	struct net_device *ret = vlan_dev_priv(dev)->real_dev;
 
-	while (is_vlan_dev(ret))
+	while (ret && is_vlan_dev(ret))
 		ret = vlan_dev_priv(ret)->real_dev;
 
 	return ret;

But will make all the callers of vlan_dev_real_dev() feel like they
should NULL-check the result, which is not necessary.

Isn't it better to reliably return NULL instead of a silent UAF in
this edge case? 

RDMA must be calling this helper on a vlan which was already
unregistered, can we fix RDMA instead?

RDMA holds a get on the netdev which prevents unregistration, however
unregister_vlan_dev() does:

        unregister_netdevice_queue(dev, head);
        dev_put(real_dev);

Which corrupts the still registered vlan device while it is sitting in
the queue waiting to unregister. So, it is not true that a registered
vlan device always has working vlan_dev_real_dev().

Jason